Oleg Drokin wrote:

>Hello!
>
>On Sat, Dec 14, 2002 at 01:29:19AM -0800, Tupshin Harper wrote:
>
>  
>
>>i'm getting the following error logged every 11 seconds or so:
>>
>>Dec 14 01:00:49 phylum kernel: vs-3050: wait_buffer_until_released: nobody
>>releases buffer (dev 16:01, size 4096, blocknr 2916352, count 3, list 0, 
>>state
>>0x10019, page c1172108, (UPTODATE, CLEAN, UNLOCKED)). Still waiting
>>(-1320000000) JDIRTY !JWAIT
>>Also, some processes are blocking, include ps (so I can't get a complete 
>>process list), and shutdown.
>>    
>>
>
>Can you please execute SysRq-T, decode it with ksymoops and send us the result?
>  
>
I have succeeded in reproducing this problem at will (and 
deterministically) with a SysRq enabled kernel.

I can reproduce the problem consistently by doing a cp -av of a 
directory(haven't tried other cp permutations, guessing it wouldn't make 
a difference) containing three mp3 files from one location on a 
partition to another location on the same reiserfs partition. During the 
third file, cp hangs(continues to use 80%+ of the CPU), and is 
unkillable. A few minutes later, it starts generating the same JDIRTY 
error messages that I reported before.

The attached file is a ksymoops processed version of SysRq-T after this 
problem has started to occur. There is one cp taking place which is the 
one that triggered the problem, and there is also a sync command that is 
blocking as well which I tried to do after the cp hung.

I have not tried to do any kind of reiserfs check or repair on the 
filesystem, and I can trigger this problem at any point if you would 
like me to test further.

-Tupshin